[ie/tv5unis] Fix extractors (#15477)

Closes #12662
Authored by: 0xced

yt_dlp/extractor/tv5unis.py

@@ -1,14 +1,18 @@
 from .common import InfoExtractor
 from ..utils import (
+    ExtractorError,
     int_or_none,
+    join_nonempty,
+    make_archive_id,
     parse_age_limit,
-    smuggle_url,
+    remove_end,
-    try_get,
 )
+from ..utils.traversal import traverse_obj
 
 
 class TV5UnisBaseIE(InfoExtractor):
     _GEO_COUNTRIES = ['CA']
+    _GEO_BYPASS = False
 
     def _real_extract(self, url):
         groups = self._match_valid_url(url).groups()
@@ -16,96 +20,136 @@ class TV5UnisBaseIE(InfoExtractor):
             'https://api.tv5unis.ca/graphql', groups[0], query={
                 'query': '''{
   %s(%s) {
+    title
+    summary
+    tags
+    duration
+    seasonNumber
+    episodeNumber
     collection {
       title
     }
-    episodeNumber
     rating {
       name
     }
-    seasonNumber
-    tags
-    title
     videoElement {
+      __typename
       ... on Video {
         mediaId
+        encodings {
+          hls {
+            url
+          }
+        }
+      }
+      ... on RestrictedVideo {
+        code
+        reason
       }
     }
   }
 }''' % (self._GQL_QUERY_NAME, self._gql_args(groups)),  # noqa: UP031
             })['data'][self._GQL_QUERY_NAME]
-        media_id = product['videoElement']['mediaId']
+
+        video = product['videoElement']
+        if video is None:
+            raise ExtractorError('This content is no longer available', expected=True)
+
+        if video.get('__typename') == 'RestrictedVideo':
+            code = video.get('code')
+            if code == 1001:
+                self.raise_geo_restricted(countries=self._GEO_COUNTRIES)
+            reason = video.get('reason')
+            raise ExtractorError(join_nonempty(
+                'This video is restricted',
+                code is not None and f', error code {code}',
+                reason and f': {remove_end(reason, ".")}',
+                delim=''))
+
+        media_id = video['mediaId']
+        formats, subtitles = self._extract_m3u8_formats_and_subtitles(
+            video['encodings']['hls']['url'], media_id, 'mp4')
 
         return {
-            '_type': 'url_transparent',
             'id': media_id,
-            'title': product.get('title'),
+            '_old_archive_ids': [make_archive_id('LimelightMedia', media_id)],
-            'url': smuggle_url('limelight:media:' + media_id, {'geo_countries': self._GEO_COUNTRIES}),
+            'formats': formats,
-            'age_limit': parse_age_limit(try_get(product, lambda x: x['rating']['name'])),
+            'subtitles': subtitles,
-            'tags': product.get('tags'),
+            **traverse_obj(product, {
-            'series': try_get(product, lambda x: x['collection']['title']),
+                'title': ('title', {str}),
-            'season_number': int_or_none(product.get('seasonNumber')),
+                'description': ('summary', {str}),
-            'episode_number': int_or_none(product.get('episodeNumber')),
+                'tags': ('tags', ..., {str}),
-            'ie_key': 'LimelightMedia',
+                'duration': ('duration', {int_or_none}),
+                'season_number': ('seasonNumber', {int_or_none}),
+                'episode_number': ('episodeNumber', {int_or_none}),
+                'series': ('collection', 'title', {str}),
+                'age_limit': ('rating', 'name', {parse_age_limit}),
+            }),
         }
 
 
 class TV5UnisVideoIE(TV5UnisBaseIE):
-    _WORKING = False
     IE_NAME = 'tv5unis:video'
-    _VALID_URL = r'https?://(?:www\.)?tv5unis\.ca/videos/[^/]+/(?P<id>\d+)'
+    _VALID_URL = r'https?://(?:www\.)?tv5unis\.ca/videos/[^/?#]+/(?P<id>\d+)'
-    _TEST = {
+    _TESTS = [{
-        'url': 'https://www.tv5unis.ca/videos/bande-annonces/71843',
+        'url': 'https://www.tv5unis.ca/videos/bande-annonces/144041',
-        'md5': '3d794164928bda97fb87a17e89923d9b',
+        'md5': '24a247c96119d77fe1bae8b440457dfa',
         'info_dict': {
-            'id': 'a883684aecb2486cad9bdc7bbe17f861',
+            'id': '56862325352147149dce0ae139afced6',
+            '_old_archive_ids': ['limelightmedia 56862325352147149dce0ae139afced6'],
             'ext': 'mp4',
-            'title': 'Watatatow',
+            'title': 'Antigone',
-            'duration': 10.01,
+            'description': r"re:En aidant son frère .+ dicté par l'amour et la solidarité.",
+            'duration': 61,
         },
-    }
+    }]
     _GQL_QUERY_NAME = 'productById'
 
     @staticmethod
     def _gql_args(groups):
-        return f'id: {groups}'
+        return f'id: {groups[0]}'
 
 
 class TV5UnisIE(TV5UnisBaseIE):
-    _WORKING = False
     IE_NAME = 'tv5unis'
-    _VALID_URL = r'https?://(?:www\.)?tv5unis\.ca/videos/(?P<id>[^/]+)(?:/saisons/(?P<season_number>\d+)/episodes/(?P<episode_number>\d+))?/?(?:[?#&]|$)'
+    _VALID_URL = r'https?://(?:www\.)?tv5unis\.ca/videos/(?P<id>[^/?#]+)(?:/saisons/(?P<season_number>\d+)/episodes/(?P<episode_number>\d+))?/?(?:[?#&]|$)'
     _TESTS = [{
-        'url': 'https://www.tv5unis.ca/videos/watatatow/saisons/6/episodes/1',
+        # geo-restricted to Canada; xff is ineffective
-        'md5': 'a479907d2e531a73e1f8dc48d6388d02',
+        'url': 'https://www.tv5unis.ca/videos/watatatow/saisons/11/episodes/1',
+        'md5': '43beebd47eefb1c5caf9a47a3fc35589',
         'info_dict': {
-            'id': 'e5ee23a586c44612a56aad61accf16ef',
+            'id': '2c06e4af20f0417b86c2536825287690',
+            '_old_archive_ids': ['limelightmedia 2c06e4af20f0417b86c2536825287690'],
             'ext': 'mp4',
-            'title': 'Je ne peux pas lui résister',
+            'title': "L'homme éléphant",
-            'description': "Atys, le nouveau concierge de l'école, a réussi à ébranler la confiance de Mado en affirmant qu'une médaille, ce n'est que du métal. Comme Mado essaie de lui prouver que ses valeurs sont solides, il veut la mettre à l'épreuve...",
+            'description': r're:Paul-André et Jean-Yves, .+ quand elle parle du feu au Spot.',
             'subtitles': {
                 'fr': 'count:1',
             },
-            'duration': 1370,
+            'duration': 1440,
             'age_limit': 8,
-            'tags': 'count:3',
+            'tags': 'count:4',
             'series': 'Watatatow',
-            'season_number': 6,
+            'season': 'Season 11',
+            'season_number': 11,
+            'episode': 'Episode 1',
             'episode_number': 1,
         },
     }, {
-        'url': 'https://www.tv5unis.ca/videos/le-voyage-de-fanny',
+        # geo-restricted to Canada; xff is ineffective
-        'md5': '9ca80ebb575c681d10cae1adff3d4774',
+        'url': 'https://www.tv5unis.ca/videos/boite-a-savon',
+        'md5': '7898e868e8c540f03844660e0aab6bbe',
         'info_dict': {
-            'id': '726188eefe094d8faefb13381d42bc06',
+            'id': '4de6d0c6467b4511a0c04b92037a9f15',
+            '_old_archive_ids': ['limelightmedia 4de6d0c6467b4511a0c04b92037a9f15'],
             'ext': 'mp4',
-            'title': 'Le voyage de Fanny',
+            'title': 'Boîte à savon',
-            'description': "Fanny, 12 ans, cachée dans un foyer loin de ses parents, s'occupe de ses deux soeurs. Devant fuir, Fanny prend la tête d'un groupe de huit enfants et s'engage dans un dangereux périple à travers la France occupée pour rejoindre la frontière suisse.",
+            'description': r're:Dans le petit village de Broche-à-foin, .+ celle qui fait battre son coeur.',
             'subtitles': {
                 'fr': 'count:1',
             },
-            'duration': 5587.034,
+            'duration': 1200,
-            'tags': 'count:4',
+            'tags': 'count:5',
         },
     }]
     _GQL_QUERY_NAME = 'productByRootProductSlug'